Question 1

What problem does Kubernetes solve?

Accepted Answer

Kubernetes helps teams deploy, scale, manage, and recover containerized applications across a cluster of machines.

It solves common operational problems such as service discovery, self-healing, rollout management, configuration injection, and horizontal scaling. In interviews, good answers connect Kubernetes to running distributed systems reliably rather than describing it as “just a container tool.”

Question 2

What is a Pod in Kubernetes?

Accepted Answer

A Pod is the smallest deployable unit in Kubernetes and usually contains one main application container, though it can contain multiple tightly coupled containers.

Pods are ephemeral, so they should not be treated like long-lived pets. Kubernetes expects higher-level controllers to create and replace Pods as needed.

Question 3

What is the difference between a Deployment and a StatefulSet?

Accepted Answer

A Deployment is typically used for stateless applications and manages replica updates, rollout history, and scaling. A StatefulSet is used for workloads that need stable identities, ordered deployment, and persistent storage behavior.

A strong answer should mention that StatefulSets are common for databases and clustered systems, while Deployments are more common for web apps, APIs, and background workers.

Question 4

What is a Kubernetes Service?

Accepted Answer

A Service provides stable networking and a consistent way to reach a set of Pods selected by labels, even when the individual Pod IPs change.

This is important because Pods come and go. Services give applications a stable abstraction for internal communication and load balancing inside the cluster.

Question 5

What are readiness and liveness probes?

Accepted Answer

Readiness probes determine whether a Pod should receive traffic. Liveness probes determine whether a container is stuck or unhealthy and should be restarted.

These probes are important because they help Kubernetes make better decisions about traffic routing and recovery. A weak probe configuration can cause flapping, false restarts, or traffic being sent to broken instances.

Question 6

How do ConfigMaps and Secrets differ?

Accepted Answer

Both are used to inject configuration into workloads, but Secrets are intended for sensitive values while ConfigMaps are for non-sensitive configuration.

In practice, the distinction matters operationally and from a security perspective. Good answers may also mention that Secrets still require proper cluster security and should not be treated as magically safe just because they use a different Kubernetes object type.

Question 7

What is an Ingress?

Accepted Answer

Ingress is an API object that defines HTTP and HTTPS routing rules to services inside the cluster, typically managed by an ingress controller.

It provides a structured way to expose multiple services under shared domains, TLS settings, and path-based routing. A good answer should mention that Ingress needs a controller to actually enforce the rules.

Question 8

How does the Horizontal Pod Autoscaler work?

Accepted Answer

The Horizontal Pod Autoscaler, or HPA, adjusts the number of Pod replicas based on observed metrics such as CPU, memory, or custom metrics.

It is useful when demand changes over time, but good autoscaling depends on meaningful metrics and applications that can scale horizontally without breaking state or startup assumptions.

Question 9

What is the difference between requests and limits in Kubernetes?

Accepted Answer

Resource requests define the amount of CPU and memory a container asks for when being scheduled. Limits define the maximum amount it is allowed to consume.

This affects both scheduling and runtime behavior. For example, CPU can be throttled when it exceeds the limit, while memory pressure may lead to OOM kills. A strong answer should show that you understand how bad resource settings affect stability.

Question 10

How do rolling updates work in Kubernetes?

Accepted Answer

With rolling updates, Kubernetes gradually replaces old Pods with new ones rather than stopping everything at once.

This helps reduce downtime and rollout risk. However, it works well only when readiness probes, startup behavior, and backward compatibility are configured correctly.

Question 11

How do you troubleshoot CrashLoopBackOff?

Accepted Answer

I would start with `kubectl describe pod` to inspect events, then use `kubectl logs` and `kubectl logs --previous` to see current and prior container output. After that, I would check startup commands, environment variables, probes, missing dependencies, and resource limits.

CrashLoopBackOff is a symptom, not a root cause. A good interview answer should show a structured approach to narrowing down why the container keeps failing or being restarted.

Question 12

What are common causes of production issues in Kubernetes?

Accepted Answer

Common causes include bad resource settings, broken readiness or liveness probes, failed image pulls, DNS issues, misconfigured ingress, missing secrets, storage problems, and deployment changes that are not backward compatible.

The deeper point is that Kubernetes adds an orchestration layer, so debugging often requires checking both application behavior and cluster behavior together.

Question 13

What is the control plane in Kubernetes?

Accepted Answer

The control plane is the set of components that manage cluster state, scheduling, API access, and reconciliation.

It is important because Kubernetes works by constantly comparing desired state to actual state. The control plane is what makes that model function.

Question 14

What does the scheduler do?

Accepted Answer

The scheduler decides which node should run a newly created Pod based on available resources, constraints, and policies.

A good answer should mention that scheduling is influenced by requests, taints, tolerations, affinity rules, and other placement controls.

Question 15

What is etcd in Kubernetes?

Accepted Answer

`etcd` is the distributed key-value store that holds cluster state and configuration for the Kubernetes control plane.

It is a critical component because losing or corrupting it can affect the whole cluster. Mature operations treat etcd backup and protection seriously.

Question 16

What are labels and selectors used for?

Accepted Answer

Labels are key-value metadata attached to Kubernetes objects. Selectors use those labels to find matching resources.

They are fundamental to how Services, Deployments, and many other controllers associate related objects. Good labeling strategy improves manageability and clarity.

Question 17

What is a namespace?

Accepted Answer

A namespace is a logical partition within a cluster used to organize and isolate resources.

Namespaces help separate teams, applications, or environments, but they are not a complete security boundary by themselves. Strong answers often mention RBAC alongside namespaces.

Question 18

What is RBAC in Kubernetes?

Accepted Answer

RBAC, or Role-Based Access Control, defines which users or service accounts can perform which actions on which resources in the cluster.

It is a key security control because many incidents come from excessive cluster permissions. Good answers mention least privilege and scoped roles.

Question 19

What is a DaemonSet?

Accepted Answer

A DaemonSet ensures that a Pod runs on every node, or every matching node, in the cluster.

This is useful for node-level agents such as log collectors, monitoring agents, or security tooling. It shows how Kubernetes supports cluster-wide operational services.

Question 20

What is a Job or CronJob in Kubernetes?

Accepted Answer

A Job runs a task to completion, while a CronJob runs Jobs on a schedule.

These resources are useful for batch processing, scheduled maintenance, and periodic operational tasks. They are better fits than Deployments for workloads that should not run continuously.

Question 21

What is persistent storage in Kubernetes?

Accepted Answer

Persistent storage allows data to survive Pod restarts and rescheduling. It is commonly managed using PersistentVolumes and PersistentVolumeClaims.

This matters because Pods are ephemeral, but many real workloads need durable data. Good answers should mention that storage design depends on the application’s state model.

Question 22

What are taints and tolerations?

Accepted Answer

Taints repel Pods from nodes unless the Pods explicitly tolerate them. They are used to control which workloads can run on particular nodes.

This is useful for dedicating nodes to special workloads such as GPUs, critical services, or infrastructure components. It is part of more advanced scheduling control.

Question 23

What is node affinity?

Accepted Answer

Node affinity lets you express preferences or requirements for which nodes a Pod should run on based on labels.

It helps with workload placement when certain applications need specific hardware, zones, or isolation characteristics. A good answer may compare it to taints and tolerations.

Question 24

What is an init container?

Accepted Answer

An init container runs before the main application container starts and is often used for setup tasks such as waiting for dependencies or preparing files.

It is useful because it separates initialization logic from the main app runtime. This often leads to cleaner container design.

Question 25

How do you debug a pod that is Pending?

Accepted Answer

I would check `kubectl describe pod` for scheduling events, then inspect resource requests, node capacity, taints, affinity rules, missing PVCs, or image pull restrictions.

Pending usually means the Pod has not been placed or cannot start due to unmet cluster requirements. It is a scheduling or dependency issue, not the same as a runtime crash.

Question 26

What is a service mesh at a high level?

Accepted Answer

A service mesh adds a dedicated layer for service-to-service communication concerns such as traffic routing, retries, mutual TLS, and observability.

It can add powerful features, but it also adds complexity. A mature answer explains both the benefits and operational overhead.

Question 27

How do you make Kubernetes safer in production?

Accepted Answer

I would use namespaces, RBAC, resource requests and limits, network policies, secure image practices, observability, and careful rollout controls.

Production safety also depends on platform discipline: backup strategy, upgrade planning, and clear ownership across clusters and applications.

Question 28

What is a network policy in Kubernetes?

Accepted Answer

A network policy defines which Pods are allowed to communicate with which other Pods or endpoints.

It is useful for limiting east-west traffic and reducing blast radius inside the cluster. Good answers mention that enforcement depends on the cluster network plugin.

Question 29

Why are cluster upgrades important to plan carefully?

Accepted Answer

Cluster upgrades can affect API compatibility, workloads, controllers, and operational tooling, so they should be tested and sequenced carefully.

Strong answers mention version skew, deprecations, and validating critical workloads before and after upgrade.

Question 30

What is an admission controller?

Accepted Answer

An admission controller intercepts API requests and can validate or mutate resources before they are stored in the cluster.

It is commonly used for security, policy, and governance controls. This is an advanced but valuable Kubernetes concept.

Container Orchestration (Kubernetes)

1. What problem does Kubernetes solve?

2. What is a Pod in Kubernetes?

3. What is the difference between a Deployment and a StatefulSet?

4. What is a Kubernetes Service?

5. What are readiness and liveness probes?

6. How do ConfigMaps and Secrets differ?

7. What is an Ingress?

8. How does the Horizontal Pod Autoscaler work?

9. What is the difference between requests and limits in Kubernetes?

10. How do rolling updates work in Kubernetes?

11. How do you troubleshoot CrashLoopBackOff?

12. What are common causes of production issues in Kubernetes?

13. What is the control plane in Kubernetes?

14. What does the scheduler do?

15. What is etcd in Kubernetes?

16. What are labels and selectors used for?

17. What is a namespace?

18. What is RBAC in Kubernetes?

19. What is a DaemonSet?

20. What is a Job or CronJob in Kubernetes?

21. What is persistent storage in Kubernetes?

22. What are taints and tolerations?

23. What is node affinity?

24. What is an init container?

25. How do you debug a pod that is Pending?

26. What is a service mesh at a high level?

27. How do you make Kubernetes safer in production?

28. What is a network policy in Kubernetes?

29. Why are cluster upgrades important to plan carefully?

30. What is an admission controller?